The European Commission has published the first draft of the Code of Practice on marking and labelling AI-generated content , marking a key step toward implementing the transparency obligations set out in the EU’s AI Act. The publication follows the timeline for finalising the Code by June 2026.

Under Article 50 of the AI Act, providers of AI systems will be required to mark AI-generated or manipulated content in a machine-readable format, while users deploying generative AI for professional purposes must clearly label deepfakes and AI-generated text related to matters of public interest. To support compliance, the Commission is facilitating the development of a voluntary Code of Practice drafted by independent experts, ahead of the rules becoming applicable.

The draft Code is structured in two parts. The first sets out rules on marking and detecting AI-generated content and applies to providers of generative AI systems. The second focuses on the labelling of deepfakes and certain AI-generated or manipulated texts on matters of public interest, and applies to deployers of such systems.

Stakeholders participating in or observing the process can submit feedback on the first draft until 23 January. A second draft is expected by mid-March 2026, with the final Code of Practice scheduled for adoption in June 2026.

The transparency obligations for AI-generated content under the AI Act will start to apply on 2 August 2026.